Re: mm: fault in __do_fault

[Andrea Arcangeli <aarcange@redhat.com>]

Hello Dmitry,

On Tue, Feb 28, 2017 at 03:04:53PM +0100, Dmitry Vyukov wrote:
> Hello,
> 
> The following program triggers GPF in __do_fault:
> https://gist.githubusercontent.com/dvyukov/27345737fca18d92ef761e7fa08aec9b/raw/d99d02511d0bf9a8d6f6bd9c79d373a26924e974/gistfile1.txt

Can you verify this fix:

>From a65381bc86d2963713b6a9c4a73cded7dd184282 Mon Sep 17 00:00:00 2001
From: Andrea Arcangeli <aarcange@redhat.com>
Date: Tue, 28 Feb 2017 16:36:59 +0100
Subject: [PATCH 1/1] userfaultfd: shmem: __do_fault requires VM_FAULT_NOPAGE

__do_fault assumes vmf->page has been initialized and is valid if
VM_FAULT_NOPAGE is not returned by vma->vm_ops->fault(vma, vmf).

handle_userfault() in turn should return VM_FAULT_NOPAGE if it doesn't
return VM_FAULT_SIGBUS or VM_FAULT_RETRY (the other two
possibilities).

This VM_FAULT_NOPAGE case is only invoked when signal are pending and
it didn't matter for anonymous memory before. It only started to
matter since shmem was introduced. hugetlbfs also takes a different
path and doesn't exercise __do_fault.

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
---
 fs/userfaultfd.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/userfaultfd.c b/fs/userfaultfd.c
index fb6d02b..de28f43 100644
--- a/fs/userfaultfd.c
+++ b/fs/userfaultfd.c
@@ -500,7 +500,7 @@ int handle_userfault(struct vm_fault *vmf, unsigned long reason)
 			 * in such case.
 			 */
 			down_read(&mm->mmap_sem);
-			ret = 0;
+			ret = VM_FAULT_NOPAGE;
 		}
 	}